Currently the different conditions are evaluated and theirs output just summed up for all activities.
This sum should be replaced by a neural network with an input node per condition and an output node per activity to calculate the current likelihood for each.
Or one network per activity?
=> a design to be done first...
Currently the different conditions are evaluated and theirs output just summed up for all activities.
This sum should be replaced by a neural network with an input node per condition and an output node per activity to calculate the current likelihood for each.
Or one network per activity?
=> a design to be done first...